AI-based real-time energy management system for high energy efficiency in logistics centers based on accuracy of forecast data

1. An energy management system for enhancing energy efficiency in a logistics center based on accuracy of forecast data, comprising: a data collection server configured to collect data related to energy management of the logistics center; and an energy management server configured to manage energy within the logistics center using the collected data, wherein the energy management server is configured to: predict future power demand and future external temperature by using historical external temperature data and historical power usage data included in the collected data, determine a difference value of the future power demand by using a difference between the predicted external temperature at a current point in time and an actual external temperature included in the collected data, predict power demand by correcting the future power demand using the difference value of the future power demand, input time-specific cloud forecasts included in the collected data into a pre-trained AI model to cluster time-specific solar radiation, and predict solar power generation based on the clustered time-specific solar radiation, calculate prediction accuracy by comparing the predicted power demand and the predicted solar power generation with the collected data, update a first peak power to a second peak power by applying the prediction accuracy to the first peak power, calculate a Mean Squared Error (MSE) between the clustered time-specific solar radiation and actual solar radiation, calculate accuracy of solar power generation prediction based on comparison of the MSE with a preset reference value, and determine the second peak power based on the first peak power, accuracy of power demand prediction, the accuracy of the solar power generation prediction, and a preset ratio constant.

2. The energy management system according to claim 1, wherein the energy management server is configured to calculate the accuracy of the power demand prediction by using a ratio between a corrected value of the future power demand and the predicted future power demand.

3. The energy management system according to claim 1, wherein the energy management server is configured to predict the future power demand and the future external temperature by inputting the historical external temperature data and the historical power usage data into at least one pre-trained time-series-based AI model.

4. The energy management system according to claim 1, wherein the energy management server is configured to optimize charging and discharging scheduling of an Energy Storage System (ESS) within the logistics center based on a preset scheduling optimization criterion, wherein the scheduling optimization criterion comprises at least one of: a first criterion that a current state of charge is similar to a previous state of charge, a second criterion of preventing complete discharge and complete charge, a third criterion of minimizing energy usage cost, and a fourth criterion that total power usage is less than or equal to the second peak power.

5. The energy management system according to claim 4, wherein the energy management server is configured to optimize operation of an HVAC system within the logistics center based on a preset operation optimization criterion when the scheduling optimization criterion fails to meet the fourth criterion, wherein the operation optimization criterion comprises at least one of: a fifth criterion that the total power used within the logistics center is less than the second peak power, and a sixth criterion that set temperature of the HVAC system satisfies a preset temperature condition.

6. An energy management method for enhancing energy efficiency in a logistics center, comprising: predicting future power demand and future external temperature by using historical external temperature data and historical power usage data included in collected data, determining a difference value of the future power demand by using a difference between the predicted external temperature at a current point in time and an actual external temperature included in the collected data, predicting power demand by correcting the future power demand using the difference value of the future power demand, inputting time-specific cloud forecasts included in the collected data into a pre-trained AI model to cluster time-specific solar radiation and predict solar power generation based on the clustered time-specific solar radiation, calculating prediction accuracy by comparing the predicted power demand and the predicted solar power generation with the collected data, updating a first peak power to a second peak power by applying the prediction accuracy to the first peak power, wherein the predicting solar power generation comprises: calculating a Mean Squared Error (MSE) between the clustered time-specific solar radiation and actual solar radiation, and calculating accuracy of solar power generation prediction based on comparison of the MSE with a preset reference value, and wherein the updating a first peak power to a second peak power comprises: determining the second peak power based on the first peak power, accuracy of power demand prediction, the accuracy of the solar power generation prediction, and a preset ratio constant.

7. The energy management method according to claim 6, further comprising: calculating the accuracy of the power demand prediction by using a ratio between a corrected value of the future power demand and the predicted future power demand.

8. The energy management method according to claim 6, further comprising: predicting the future power demand and the future external temperature by inputting the historical external temperature data and the historical power usage data into at least one pre-trained time-series-based AI model.